I don't think this addresses the primary concern about associative reads which is the house of cards thing. You asked me about my W/W read on you and LC - what if LC flips town? Similarly I'll have to ask you here - what if nanook flips scum? You haven't solved the main flaw with associative reads which you seem to be so concerned about. That aside...JaggedJimmyJay wrote: ↑Thu Jul 09, 2020 5:36 pmHypocrisy is a staple of the civilian playbook.![]()
I have made numerous associative reads. I literally charted them. I may use them a bit differently than you do, however. I use them as a part of the POE process (at least when they precede flips). Determining what teams don't work helps to eliminate suspects.
I don't stick these meaningless numbers on my reads. I said it's nice for Epi, and it is (to whatever degree you wish to describe it as "nice"). It doesn't clear Epi or mean that I'm finished assessing Epi. It's a single data point that I extracted in a review and posted. It was the most visible. There may be other connections of interest to uncover, but they're going to require a more dedicated analytic approach. I did not notice them in a general ISO scan. I have a very specific methodology for looking into two-way pairings as a deliberate motion. When I do so for Epi, we'll see what I find.Alison wrote: ↑Thu Jul 09, 2020 5:29 pmbut also this logic doesn't even make sense. Like let's say we think that Epi has 20% chance to be partnered with nanook if he is indeed scum and 15% chance to be partnered with the next-highest percent. Why can't we be living in that 15% world? It's not that unlikely. For the chances of Epi/nanook not being the case to significantly affect your read on Epi, you have to really be sure that Epi is very likely to be scum with nanook if he is scum. Like 60/20 instead of 20/15. Then maybe it makes sense.
...the numbers are examples, but they aren't meaningless. If we take "Epi is town" to be A, "Epi and nanook are scum" to be B, and "Epi is scum but not with nanook" to be C, then it is incorrect to believe that not-B strongly implies A unless you also believe C to be very low. For instance, if B = 0% (if we take it on faith that Epi/nanook are never a team together) and C = 80% (Epi is individually scummy, so he's scum wih someone that isn't nanook, since we know B is 0), then A = 20% (Epi has a low chance of being town). In other words, the probability of Epi/nanook should not significanatly affect your read on Epi unless you think Epi can't be partnered with anyone other than nanook. This suggests that the single data point B should not significantly adjust your read of A (since I have proven that A can be very high even if B is very low), which is what your post sounded like it was doing.